Artigo e projeto de investigação
RESEARCH INTEGRITY: PREVALENCE OF
RESEARCH MISCONDUCT, SEVERITY
ATTRIBUTED AND CONDITIONING REASONS
Aluno
CLÁUDIA FILIPA REIS MACHADO DA SILVA
Mestrado Integrado em Medicina (MIM) do ICBAS/UP e HSA/CHP
Orientadora:
Prof. Doutora Margarida Lima
HSA/CHP e ICBAS/UP
O artigo de investigação original, apresentado para fins de obtenção do grau de Mestre em Medicina, resultou de um projeto de investigação desenvolvido no âmbito da Disciplina de Iniciação à Investigação Clínica (DIIC) do Curso de Mestrado Integrado em Medicina (MIM) do Instituto de Ciências Biomédicas Abel Salazar da Universidade do Porto (ICBAS/UP) e do Centro Hospitalar do Porto (CHP) e vai ser submetido para publicação em revista científica da área.
A proposta de projeto de investigação subjacente a este artigo, que se apresenta no final do mesmo, foi elaborada durante o ano letivo 2012/2013 e o projeto foi executado durante o ano letivo 2013/2014, sob a orientação e supervisão da Prof. Doutora Margarida Lima, docente responsável pela DIIC, com colaboração da Dra. Isabel Fonseca, colaboradora do Gabinete Coordenador da Investigação do Departamento de Ensino, Formação e Investigação do CHP.
1. ARTIGO DE INVESTIGAÇÃO... 6
Research integrity: Prevalence of research misconduct, severity attributed and conditioning reasons ... 6
Abstract ... 8
Introduction ... 9
Materials and Methods ... 11
Results and Discussion ... 12
Sample characterization ... 12
Prevalence of research misconduct ... 14
Severity attributed to research misconduct practices ... 19
Conditioning reasons for research misconduct ... 22
Limitations ... 23 Conclusions ... 23 Acknowledgments ... 24 Author Contributions ... 24 Competing Interests ... 24 References ... 24
2. PROPOSTA DE PROJETO DE INVESTIGAÇÃO... 26
Integridade em Investigação ... 26 Plano científico ... 27 Resumo ... 28 Introdução ... 29 Problemas ... 35 Questões ... 35 Objetivos do estudo ... 35 Intervenientes ... 36
Instituições, Departamentos e Serviços ... 36
Equipa de Investigação ... 36
Metodologia ... 38
Critérios de revisão da literatura ... 38
Material e métodos ... 39
Análise de dados ... 39
Plano de trabalho ... 39
Calendarização ... 41
Duração ... 41
Datas de início e conclusão ... 41
Cronograma ... 41
Indicadores de produção ... 42
Comunicações orais e posters ... 42
Trabalhos escritos ... 42
Referências bibliográficas ... 43
Questões éticas ... 46
Informação dos participantes ... 47
Confidencialidade e anonimização ... 47 Plano financeiro... 48 Orçamento ... 49 Financiamento ... 49 Glossário ... 50 Siglas e acrónimos ... 51 Anexos ... 52
Anexo 1 – Proposta de questionário ... 54
Anexo 2 – Folha de avaliação da proposta de questionário ... 58
Anexo 3 – Resultados da avaliação da proposta de questionário ... 60
Anexo 4 – Versão eletrónica do questionário. ... 61
Anexo 5 – Informação a ser fornecida aos participantes ... 66
Adendas ... 67
Folha de rosto do estudo de investigação ... 69
Pedidos de autorização institucional ... 70
Presidente do Conselho de Administração do CHP ... 71
Presidente da Comissão de Ética para a Saúde do CHP ... 71
Termos de responsabilidade ... 72
Aluno ... 72
Orientador do projeto ... 72
Supervisor do projeto / Responsável pela DIIC ... 72
Termos de autorização local ... 73
Diretora do Departamento de Ensino, Formação e Investigação ... 73
1. ARTIGO DE INVESTIGAÇÃO
Research integrity: Prevalence of research misconduct, severity
attributed and conditioning reasons
RESEARCH INTEGRITY: PREVALENCE OF RESEARCH MISCONDUCT, SEVERITY ATTRIBUTED AND CONDITIONING REASONS
Authors: Cláudia Silva1, Isabel Fonseca2, Margarida Lima3
1 Medical student, Instituto de Ciências Biomédicas de Abel Salazar, Universidade do Porto (ICBAS/UP), Porto, Portugal.
2 RD, MSc, Hospital de Santo António (HSA), Centro Hospitalar do Porto (CHP); Unidade Multidisciplinar de Investigação Biomédica (UMIB), Porto, Portugal.
3 MD, PhD, Hospital de Santo António (HSA), Centro Hospitalar do Porto (CHP); Instituto de Ciências Biomédicas Abel Salazar, Universidade do Porto (ICBAS/UP); Unidade Multidisciplinar de Investigação Biomédica (UMIB), Porto, Portugal.
Cláudia Silva
Complete name: Cláudia Filipa Reis Machado da Silva Academic degree: Graduated in Biomedical Sciences ORCID number: 0000-0002-3998-8058
Personal e-mail: claudiasilva120@gmail.com
Isabel Fonseca
Complete name: Isabel Maria da Silva Fonseca Academic degree: RD, MSc
ORCID number: 0000-0001-8984-1751
Professional e-mail: ifonseca.defi@chporto.min-saude.pt
Margarida Lima (corresponding author)
Complete name: Margarida Maria de Carvalho Lima Academic degree: PhD
ORCID number: 0000-0001-9702-5260
Professional e-mail: margaridalima@chporto.min-saude.pt Personal e-mail: margaridamariacarvalholima@gmail.com
Postal address: Serviço de Hematologia, Hospital de Santo António, Centro Hospitalar do Porto, Rua D. Manuel II, s/n. 4099-001, Porto, Portugal.
Telephone number: +351-22-2077500
Abstract
Background: Research misconduct has been under increasing attention throughout
scientific community in the last decades, however its definition is not yet consensual, its prevalence is not still precisely known and its reasons are poorly characterized.
Objectives: To known the prevalence of misconduct practices among the researchers from
biomedical institutions, its recognition in other cases, the severity attached to different situations and their conditioning reasons.
Methods: One hundred and seventy investigators from five Portuguese biomedical
institutions responded voluntarily to a questionnaire developed by the authors.
Results: The most prevalent admitted situation was citation without consult the whole
article. On the other hand, the most frequently observed practice was inappropriate inclusion of investigators in a research team. The two areas with larger proportion of malpractices, as well as more observed cases, were “planning and execution of research proposals” and “composition of the research team”. The practice to which researchers attributed greater severity was data fabrication or falsification. The reason with higher impact on researchers’ activity was lack of time, but when asked about the research practice in general, they considered the need to increase the number of publications
Conclusion: This study contributes to a better understanding of the extent of research
misconduct, to the perception of which malpractices are considered more serious by researchers, as well as their conditioning reasons, allowing for a better evaluation of the problem and making possible to implement preventive measures.
Introduction
Research is the major engine for scientific progress, resulting in the publication of many articles annually. For a long time, it was assumed that this activity was sufficiently self-regulated and bad practices were very rare. However, since 1980, with the detection of some misconduct cases, this conviction was shaken.[1, 2] Thereafter, research integrity and misconduct have been issues with increasing attention and discussion throughout the scientific community.[3-5]
It is important to set and adopt common definitions and classifications in order to assess, respond and change the research behaviours uniformly.[3] After a long debate, the research conducts have been classified into 3 general categories: deliberate misconduct (commonly defined as fabrication, falsification, and plagiarism), questionable research practices and responsible conduct of research.[3] However, there still isn't a consensual and sharp definition of which situations may be considered as research misconduct and there have been several definitions of research misconduct proposed by different organizations, none of them proved totally fitting.[6, 7]
A responsible research conduct can be defined as "conducting research in ways that fulfil the professional responsibilities of researchers, as defined by their professional organizations, the institutions for which they work and, when relevant, the government and public”.[3] On the other hand, serious research misconducts might be explained as “fabrication, falsification, plagiarism, and other practices that seriously deviate from accepted standards when proposing, conducting and reporting research to include misconduct occurring in connection with reviewing research”.[4] Furthermore, the 2000 Committee on Publication Ethics report also considers as Serious Research
Misconduct: non-obtainment of approval by the local ethics committee, concealment of data or
conflicting results, inclusion of information regarding side effects in clinical trials, non-obtainment of informed consent in human research, development of post hoc analysis without reporting them as such, honorary or ghost authorship, redundant publication, not reporting conflict of interests, non-publication of a concluded research and beginning a new one without a proper search. [6] Between serious misconducts and responsible conduct, with a more difficult definition to set, lay the questionable research conducts which can be defined as “actions that violate traditional values of the research enterprise and that may be detrimental to the research process” but do not “directly damage the integrity of the research process”.[3]
Data from misconduct practice can result from reported and investigated cases, self-report type questionnaires and review of the reasons for articles retraction. However, none of these methods is accurate enough in this evaluation.[8]
Several surveys have been conducted in order to identify research misconduct prevalence. One of these was conducted by Martinson, in 2005, and it was directed to researchers funded by the
National Institutes of Health.[9] The results on admitted practices were: 27.5% inappropriate record of data or results, 15.3% use of inappropriate research designs, 12.5% omission of the use of scarce data or unreliable conclusions of other researchers, 10% inappropriate acceptance of an authorship, 6% omission of data that contradicts finding from previous studies, 4.7% publication of the same data/results in two or more publications, 1.4% plagiarism of ideas and 0.3% falsification or fabrication of data.[9] Finally, in 2009, appeared the first meta-analysis on this topic, which concluded that, on average, 2% of researchers admit to have fabricated, falsified or modified data or results at least once, and 33.7% admitted other questionable research practices.[10] Regarding the observed behaviours in colleagues, fabrication, falsification and modification were witnessed by 14% of respondents and other questionable practices by 72%.[10]
Additionally, many studies have been reflecting concerns about proper attribution of an article authorship.[11] According to the International Committee of Medical Journal Editors (ICMJE), authors must meet all of the following three conditions: “1) substantial contributions to conception and design, acquisition of data, or analysis and interpretation of data, 2) drafting the article or revising it critically for important intellectual content, and 3) final approval of the version to be published.”.[12] Nevertheless, it was estimated that the percentage of honorary authorship (inclusion of an author that doesn’t meet ICMJE criteria) varies between 9 and 60%, and the ghost authorship (not inclusion of an author that meets ICMJE criteria) varies between 9 and 11%.[3]
It has been estimated that the percentage of retracted articles corresponds to only 0.02% of all publications, but between 1999 and 2009 a 10-fold increase in the annual retraction rate was observed.[11, 13-15] It was also found that, between 2000 and 2010, the number of fraud derived retractions increased as well.[11] Between 1928 and 2011, the main reasons for misconduct related removals from the 42 major bibliographic databases, were due to publication misconduct in 47% of cases, investigation misconduct in 20% and questionable data or results interpretation in 42%.[15] In a 2012 study, Stretton et al analyzed the publications retracted from MEDLINE, between 1966 and 2008, and found that, regarding the articles retracted because of misconduct, 52.1% were related to fabrication and falsification and 41.8% to plagiarism.[16]
After any article is retracted because of error or fraud, the journal should publish a retraction note.[12] However, Steen et al ascertained that, in 2011, 31.8 % of withdrawals weren’t noticed in any way and after retraction they can continue to influence scientific community.[17] In 1999, Budd
et al found that the retracted articles from MEDLINE between 1966 and 1996 were cited about
2000 times after withdrawal and only 8% of the citations referred this extraction.[18] The same author analyzed the articles taken out from the same database between 1997 and 2009 and concluded that the number of citations after withdrawal was similar and that the reference to the extraction was only present in 4% of cases.[19]
In Portugal, there aren’t, at least that we know of, comprehensive studies on research misconduct. This lack of knowledge restrains the implementation of preventive measures.
In this study, we propose to evaluate the prevalence of research misconduct in five Portuguese biomedical institutions, and its recognition in other cases, to assess the severity attached to different situations of research misconduct, as well as their conditioning reasons, using a questionnaire directed to their researchers.
Materials and Methods
This is a multicenter, descriptive, epidemiological and transversal study. Nine Portuguese biomedical research institutions were invited and five agreed to participate: three faculties of the University of Porto (Institute for Biomedical Sciences Abel Salazar, ICBAS/UP, Faculty of Pharmacy, FFUP, and Faculty of Medicine, FMUP) and two university hospitals (Centro Hospitalar do Porto, CHP, and Centro Hospitalar de São João, CHSJ, Porto). The study was approved by the Ethical Committee and authorized by the administration board of the institutions.
The study sample consists of 170 investigators from these institutions, which agreed on answering anonymously a questionnaire elaborated by the research team.
The questionnaire was composed by three parts: 1) demographic characteristics of the participants; 2) identification of misconduct practices and severity attributed to the different situations and 3) main reasons for research misconduct. In the second part, different aspects were separately analyzed: planning and execution of the research proposals, composition of the research team, data analysis and interpretation, manuscript writing, manuscript authorship and manuscript publication. Investigators were asked to indicate the frequency they practice each of the situations listed (“never, “occasionally” or “frequently”), if they know one or more cases (Yes / No) and the importance they attribute to the different situations (“not serious”, “mildly serious”, “serious”, “very serious”). In the third part, the investigators were requested to indicate, from a list of situations, what are the reasons for misconduct, in their personal case and in general (this part was only included in the electronic version).
The questionnaire was sent initially to 44 researchers for a pre-test evaluation and it was validated after analysis of the 12 received answers. Then it was sent to institutional email addresses and distributed as a printed version, from June to December of 2013.
A descriptive statistical analysis was performed using SPSS version 21, in order to characterize the sample and to determine the absolute and relative frequencies of the different response options. Continuous variables are presented as the medians and 25th-75th quartiles (IQR, interquartile range) and categorical variables as percentages.
Results and Discussion
Sample characterization
The questionnaire was answered by 170 individuals: 88 responses by the electronic way and 82 using the printed version. Thus, the study sample consisted of 170 individuals, 61.3% females and 38.7% males, with a median age of 39 years old (IQR: 30-49 years). The median time of professional activity was 13 years (IQR: 6-24 years) and median time of participation in research activities was 5 years (IQR: 1-15 years). Regarding the place of work, the majority of respondents (54.5%) worked exclusively in a hospital (Table 1). Concerning to the academic graduations, 94.7% of responders had a university graduation whereas the remaining had a polytechnic graduation, and 38.3% had a PhD or MsD (Table 1).
Table 1. Place of work and academic degrees
Percentage of cases (%)
Institution
University hospital 54.5
Research centre 1.8
Faculty 6.0
University hospital + Research centre 3.0 University hospital + Faculty 21.6 Research centre + Faculty 7.1 University hospital + Research centre+ Faculty 6.0
Academic degree
1st degree (graduation) 45.3 2nd degree (master degree) 18.8
3rd degree (doctoral degree) 19.5
MsD students 3.5
PhD students 12.9
At the hospital level, most of the respondents were physicians (66.5%) (Table 2). About 44% exercised functions related to university education and a small portion was connected to the polytechnic teaching (10%), mostly as invited professors or assistants (27.1% and 8.2%, respectively). Regarding the research career, most (17.8%) were principal investigators, 13.6% were research collaborators, and only 1.8% were coordinating investigators.
Table 2. Professional career
Professional career Percentage (%)
Hospital career Medical doctor 66.5 Nurse 4.1 Technician 14.7 Administrative 0.0 Other 2.4 Not applicable 12.4
University teaching career
Cathedra professor 2.9 Associate professor 7.6 Invited professor 12.4 Assistant 14.7 Other 6.5 Not applicable 55.9
Polytechnic teaching career
Coordinator professor 1.2 Associate professor 0.0 Invited professor 4.7 Assistant 3.5 Other 0.6 Not applicable 90.0 Research career Coordinating investigator 1.8 Principal investigator 17.8 Contributor investigator 13.6 Assistant 2.4 Research trainee 0.0 Research fellow 2.4 Other 8.3 Not applicable 53.8
With respect to participation in research activities, 34.1% referred to participate occasionally or frequently as coordinating investigator, 58.2% as principal investigator and 85.9% as contributor investigator (Table 3).
Table 3. Participation in research projects
Participation Never Occasionally Frequently Not applicable
Coordinating investigator 53.5 22.9 11.2 12.4 Principal investigator 32.4 37.6 20.6 9.4 Contributor investigator 11.8 57.1 28.8 2.4
Results are presented as percentage of cases
On the other hand, in relation to manuscript authorship, 70% had been occasionally or frequently the first author, 42.3% the last author and 78.8% other author (Table 4).
Table 4. Authorship in published articles
Authorship Never Occasionally Frequently Not applicable
First author 23.5 53.5 16.5 6.5
Last author 50.0 28.8 13.5 7.6
Other author 20.0 55.3 23.5 1.2 Results are presented as percentage of cases
The majority of respondents (85.1%) referred to work in a hospital, allowing a more focused analysis in this kind of research, unlike other studies that consider research in general. Age, time of professional activities and time of research practice had broad ranges, which indicates a heterogeneous sample and, therefore, allows a comprehensive analysis of the results.
Curiously, we find that the reported research publication rates were much higher than the reported participation in research projects ones (Tables 3 and 4).
Prevalence of research misconduct
Six aspects were separately analysed for the prevalence of research misconduct practices: planning and execution of the research proposals, composition of the research team, data analysis and interpretation, manuscript writing, manuscript authorship and manuscript publication.
During the planning and execution of research proposals, the most common misconduct practices, as well as the most witnessed ones, were: use of instruments for data collection without validation or author’s permission, and selection of trivial or already studied subjects in order to increase scientific production or to obtain faster results (Table 5). This finding can be related to the current major professional importance attributed to scientific production of every researcher, as well as to the time and human/material limitations that they face.
Table 5: Prevalence of research misconduct – Planning and execution of the research proposals
Situation
Frequency of practice (%)
Know other cases (%)
N O F NA No Yes
Selection of trivial or already studied subjects in order to increase
scientific production 62.7 26.6 1.2 9.5 37.6 62.4 Selection of trivial or already studied subjects in order to obtain
faster results 63.3 22.5 1.8 12.4 36.4 63.6 Use of others' scientific proposals without author’s consent 85.8 3.6 0.6 10.1 66.7 33.3 Use of instruments for data collection without validation or
author’s permission 58.0 29.0 5.3 7.7 36.1 63.9 Inappropriate planning of the research 78.0 8.3 3.0 10.7 64.4 35.6
Abbreviations: N, never; O, occasionally; F, frequently; NA, not applicable.
Related to the research team structure, we highlight the inappropriate inclusion of investigators, followed by acceptance to be inappropriately included in a research team, as the more frequent misconduct practices (Table 6). It is possible that this situation could be related to retribution of professional/personal favours, to avoid conflicts and to hierarchical superiority, among others.
Table 6: Prevalence of research misconduct – Composition of the research team Situation
Frequency of practice (%) Know other cases (%)
N O F NA No Yes
Inappropriate inclusion of investigators
44.7 34.1 11.8 9.4 21.1 78.9 Inappropriate exclusion of investigators
81.4 6.6 1.2 10.8 63.8 36.2 Acceptance to be inappropriately included in a research team
67.1 20.0 2.4 10.6 30.1 69.9 Request to be inappropriately included in a research team
85.2 4.1 0.0 10.7 50.6 49.4 Abbreviations: N, never; O, occasionally; F, frequently; NA, not applicable.
Within the data analysis and results interpretation section, the most frequent misconduct was the overvaluation of data to correspond to expectations, being the undervaluation or omission of data also frequently observed (Table 7). Beyond these situations, nearly half of the researchers referred to known cases of inadequate statistical analysis.
Table 7: Prevalence of research misconduct – Data analysis and interpretation
Situation
Frequency of practice (%)
Know other cases (%)
N O F NA No Yes
Data fabrication or falsification 86.5 4.7 0.0 8.8 70.6 29.4 Undervaluation or omission of data for not corresponding to
expectations 75.3 15.3 0.6 8.8 57.9 42.1 Overvaluation of data for corresponding to expectations 62.9 27.1 2.4 7.6 47.9 52.1 Inadequate statistical analysis 80.5 9.5 0.6 9.5 55.2 44.8
Abbreviations: N, never; O, occasionally; F, frequently; NA, not applicable.
In the manuscript writing process, apparently the most prevalent situation was citation without consulting the original article (Table 8). Besides, the use of others’ sentences/texts or ideas as their own and abusive self-citation were frequently recognized by researchers, which is very perturbing finding, since the plagiarism is one of the most consensual forms of severe misconduct.
Table 8: Prevalence of research misconduct – Manuscript writing
Situation
Frequency of practice (%)
Know other cases (%)
N O F NA No Yes
Use of others’ ideas as their own 86.7 4.4 0.6 8.2 55.6 44.4 Use of others’ sentences/texts as their own 81.5 8.9 1.9 7.6 49.3 50.7 Ask or pay someone to write the manuscript 86.1 3.8 1.3 8.9 68.4 31.6 Citation for non-scientific reasons 78.5 10.8 1.9 8.9 64.1 35.9 Omission of citation for non-scientific reasons 79.7 8.9 1.3 10.1 71.9 28.1 Citation without consult the original manuscript 43.0 41.8 7.0 8.2 32.0 68.0 Abusive self-citation 80.4 8.9 0.6 10.1 58.2 41.8
Abbreviations: N, never; O, occasionally; F, frequently; NA, not applicable.
About authorship attribution, it was possible to verify that the inappropriate inclusion of an author was referred to be occasionally or frequently practiced by 32.2% of investigators and 63.9% of them mentioned to know other examples among colleagues (Table 9). Furthermore, although the options “acceptance to be inappropriately considered as an author” and “request to be
inappropriately considered as an author” had a low admitted prevalence, they were frequently observed in others by the respondents (59.1% and 44.5%, respectively).
These evidences of honorary or other inadequate forms of authorship are consistent with the high percentage of publication versus participation in research projects and with inappropriate inclusion of investigators in a research team. To explain this situation we can think about retribution of professional/personal favours, hierarchical superiority, improve investigators’ curriculum vitae and to increase the probability of publication, among others.
Table 9: Prevalence of research misconduct – Manuscript authorship Situation
Frequency of practice (%) Know other cases (%)
N O F NA No Yes
Inappropriate inclusion of an author 58.2 25.9 6.3 9.5 36.1 63.9 Inappropriate exclusion of an author 86.7 3.2 0.6 9.5 65.4 34.6 Acceptance to be inappropriately considered as an author 76.6 10.8 0.6 12.0 40.9 59.1 Request to be inappropriately considered as an author 85.4 1.9 1.3 11.4 55.5 44.5
Abbreviations: N, never; O, occasionally; F, frequently; NA, not applicable.
The more common situation occurring during article publication was publication of the same article in more than one scientific journal (the whole article or with a few alterations) (Table 10). This is considered a bad practice because it violates the copyright law, leads to waste of editors’ time, improperly increases the amount of published literature, emphasizes the importance of the findings and may influence the meta-analysis.[20]
Table 10 Prevalence of research misconduct – Manuscript publication
Situation Frequency of practice (%) Know other cases (%) N O F NA No Yes
Manuscript submission with inconsistent results 82.3 7.0 0.6 10.1 68.2 31.8 Publication of the same article in more than one scientific journal (the
whole article or with few alterations) 80.4 8.2 1.3 10.1 61.9 38.1 Use of personnel/professional relationships to increase the probability
of publication or to accelerate the process 79.1 7.6 1.3 12.0 66.9 33.1 Omission of conflicts of interest 84.8 2.5 1.3 11.4 80.6 19.4
Making an overall analysis, we found that the three situations more commonly admitted as being practiced frequently were: inappropriate inclusion of investigators in a research team (11.8%), citation without consult the original article (7.0%) and inappropriate inclusion of an author (6.3%). Considering the occasional practice, the more common ones were: citation without consult the whole article (41.8%), inappropriate inclusion of investigators in a research team (34.1%) and use of instruments for data collection without validation or author’s permission (29.0%). Furthermore, combining the situations practiced frequently and occasionally, appears that the more commons were: citation without consult the whole article (48.8%), inappropriate inclusion of investigators in a research team (45.9%) and use of instruments for data collection without validation or author’s permission (34.3%). In contrast, the situations more often affirmed as never performed were: use of others’ ideas as their own (86.7%), inappropriate exclusion of an author (86.7%) and data fabrication or falsification (86.5%).
The knowledge of one or more cases of the options listed was more notable for the following situations: inappropriate inclusion of investigators in a research team (78.9%), acceptance of inappropriate inclusion in a research team (69.9%) and citation without consulting the original article (68.0%). On the other hand, the least witnessed situations were: omission of conflicts of interest disclosure (80.6%), omission of citation for non-scientific reasons (71.9%) and data fabrication or falsification (70.6%).
Overall, the two analysed sections with a higher average percentage of occasional/frequent practice, as well as with the highest percentage of recognized situations, were: "Planning and execution of the research proposal" and "Composition of the research team" (Table 11).
Table 11: Prevalence of research misconduct – Average occasional or frequent practice and observed situations according to sections
Section Average occasional / frequent
practice (%)
Average observed situations (% )
Planning and execution of the research
proposal 20.4 51.8
Composition of the research team 20.1 58.6 Data analysis and results interpretation 15.1 42.1
Manuscript writing 14.6 42.9
Manuscript authorship 12.7 50.5
Severity attributed to research misconduct practices
The malpractice considered more serious by researchers during planning and execution of a research proposal was the use of scientific proposals from other authors without author’s consent (Table 12).
Table 12: Severity attributed to research misconduct – Planning and execution of research proposals Situation
Score (% cases)
0 1 2 3
Selection of trivial or already studied subjects in order to increase scientific
production 5.5 39.3 45.4 9.8
Selection of trivial or already studied subjects in order to obtain faster results 6.2 38.9 43.8 11.1 Use of scientific proposals from other authors without author’s consent 1.3 4.4 11.9 82.5 Use of instruments for data collection without validation or author’s permission 2.4 26.8 47.6 23.2 Inappropriate planning of the research 1.2 10.6 40.4 47.8
Score: 0, not serious; 1, mildly serious; 2, serious, 3, very serious
Concerning the composition of the research team, researchers think that the most serious situation are the inappropriate exclusion of investigators and the request to be inappropriately included (Table 13).
Table 13: Severity attributed to research misconduct – Composition of the research team Situation
Score (% cases)
0 1 2 3
Inappropriate inclusion of investigators 1.2 23.6 43.6 31.5 Inappropriate exclusion of investigators 2.5 8.0 27.8 61.7 Acceptance to be inappropriately included in a research team 1.8 26.8 42.7 28.7 Request to be inappropriately included in a research team 2.5 6.8 31.5 59.3
Score: 0, not serious; 1, mildly serious; 2, serious, 3, very serious
Regarding the data analysis and results interpretation, data fabrication or falsification was considered the more serious situations, followed by undervaluation or omission of data for not corresponding to expectations, and inadequate statistical analysis (Table 14).
Table 14: Severity attributed to research misconduct – Data analysis and results interpretation Situation
Score (% cases)
0 1 2 3
Data fabrication or falsification 1.3 1.3 4.4 93.1 Undervaluation or omission of data for not corresponding to expectations 1.2 9.3 24.1 65.4 Overvaluation of data for corresponding to expectations 3.0 21.3 39.6 36.0 Inadequate statistical analysis 0.6 7.4 35.8 56.2
Score: 0, not serious; 1, mildly serious; 2, serious, 3, very serious
During manuscript writing, researches assumed that the use of others’ ideas or sentences/texts as their own and ask or pay someone to write the manuscript, are the more serious situations (Table 15). In contrast, citation without consulting the original manuscript and abusive self-citation were considered to be very serious in a low proportion of cases.
Table 15: Severity attributed to research misconduct – Manuscript writing Situation
Score (% cases)
0 1 2 3
Use of others’ ideas as their own 0.7 5.9 19.0 74.5 Use of others’ sentences/texts as their own 0.7 8.6 32.2 58.6 Ask or pay someone to write the manuscript 2.6 7.2 26.8 63.4 Citation for non-scientific reasons 4.7 28.0 36.7 30.7 Omission of citation for non-scientific reasons 3.3 15.3 45.3 36.0 Citation without consult the original manuscript 10.6 38.4 38.4 12.6 Abusive self-citation 4.6 36.4 40.4 18.5
Score: 0, not serious; 1, mildly serious; 2, serious, 3, very serious
When it comes to authorship attribution, the two situations considered to be more serious were the inappropriate exclusion of an author and the request to be inappropriately considered as an author (Table 16).
Table 16: Severity attributed to research misconduct – Manuscript authorship Situation
Score (% cases)
0 1 2 3
Inappropriate inclusion of an author 1.3 29.0 42.6 27.1 Inappropriate exclusion of an author 1.3 6.5 30.7 61.4 Accept to be inappropriately considered as an author 2.0 19.1 43.4 35.5 Request to be inappropriately considered as an author 1.3 5.9 31.4 61.4
Score: 0, not serious; 1, mildly serious; 2, serious, 3, very serious
Finally, about manuscript publication, the situations with higher classifications were: manuscript submission with inconsistent results and the use of personnel/professional relationships to increase the probability or to accelerate the process of publication (Table 17). Curiously, the omission of conflicts of interest was considered “very serious” by less than 1% of responders.
Table 17: Severity attributed to research misconduct – Manuscript publication Situation
Score (% cases)
0 1 2 3
Manuscript submission with inconsistent results 2.6 7.8 41.8 47.7 Publication of the same article in more than one scientific journal (the whole
article or with a few alterations) 1.9 16.1 42.6 39.4 Use of personnel/professional relationships to increase the probability of
publication or to accelerate the process 0.7 23.3 33.3 42.7 Omission of conflicts of interest 7.9 37.5 53.9 0.7
Score: 0, not serious; 1, mildly serious; 2, serious, 3, very serious
In overall, the most frequently situations classified as “very serious” were: data fabrication or falsification (93.1%), use of others' scientific proposals without author’s consent (82.5%) and use of others’ ideas as their own (74.5%). Considering the “serious” situations, the more often selected situations were: omission of conflicts of interest disclosure (53.9%), use of instruments for data collection without validation or author’s permission (47.6%) and selection of trivial or already studied subjects in order to increase scientific production (45.4%). Additionally, considering both “serious” and “very serious” classifications, it’s possible to see that the higher scores belongs to: data fabrication or falsification (97.5%), use of others' scientific proposals without author’s consent (94.4%) and use of others’ ideas as their own (93.5%). On the other hand, the situations more often considered as “not serious” were: citation without consult the whole article (10.6%), omission of
conflicts of interest disclosure (7.9%) and selection of trivial or already studied subjects in order to obtain faster results (6.2%).
Conditioning reasons for research misconduct
The five conditioning reasons with higher impact on their own researcher activity were: lack of time, insufficient human resources, complexity/delay in submission/approval of projects, insufficient specific formation and insufficient material/financial resources (Figure 1). These were selected on average by 43.6% of researchers. In contrast, the conviction that the risk of misconduct detection / punishment is low and the ambition for career progression or recognition / reputation were selected in a very low of cases (on average, 5.7% and 13.1%, respectively).
On the other hand, when asked about the research practice in general, and in addition to the reasons mentioned above, other reasons that they considered relevant were the need to increase the number of publications, ambition for career progression, pressure to obtain financing, pressure from scientific external evaluation and desire for recognition/reputation. On average, these reasons were selected by 76.5% of respondents. Curiously, the conviction that the risk of misconduct detection / punishment is low and the lack of clear and reliable regulation were mentioned by a relatively high proportion of the respondents (46.6 to 58.0%).
Analysing all the listed options, we found that, on average, 27.6% of researchers consider them as conditioners of their own conduct, but 71.1% believe they are conditioners of the research practice in general. 58,0 46,6 54,5 73,9 83,0 85,2 79,5 77,3 73,9 71,6 71,6 72,7 76,1 23,9 5,7 5,7 12,5 13,6 23,9 27,3 28,4 43,2 38,6 45,5 38,6 52,3 0 10 20 30 40 50 60 70 80 90 100
Lack of clear and reliable regulation Conviction that the risk of misconduct punishment is low Conviction that the risk of misconduct detection is low Desire for recognition/reputation Ambition for career progression Need to increase the number of publications Pressure to obtain financing Pressure from scientific external evaluation Complexity/ delay in submission/approval of projects Insufficient material/financial resources Insufficient human resources Insufficient specific formation Lack of time
% cases
Figure 1: Conditioning reasons for research misconduct in personal case (black bars) and in
general (gray bars). Researchers were asked, for each of the situations listed, if they consider it a reason for conditioning misconduct in their personal case and in general (yes / no).
Limitations
Although questionnaires are one of the most used methods to assess misconduct prevalence, allowing comparing numbers of admitted, observed and reported cases, they have several limitations.[2] The observation of the same case by more than one investigator may lead to an overestimation of the number of witnessed cases. As these cases are mostly not investigated, there is no certainty that it's actually a misconduct practice. Moreover, a researcher may not be aware of misconduct situations occurred in projects in which he has participated, even though he made them himself.
In this study, considering the participation of 5 different institutions, we consider the response rate extremely low, being one of the major limitations. This could be explained by indifference related to this topic, the large number of questionnaires addressed to researchers and/or the fact that the questionnaire was extensive.
Comparing these results with existing literature is hampered by the paucity of studies as comprehensive and detailed as this. Indeed, evaluated practices are different, fewer and/or more general and the classification criteria of frequency of practice vary from study to study. In addition, the results obtained in this study were based on subjective answers to the questionnaire about frequency of practice and severity attributed to the different situations. This may take some objectivity to the results, but this is a question difficult to solve in this kind of studies.
Conclusions
Despite the growing attention of the scientific community to research integrity during the last decades, there have been a few papers portraying the frequency with which researchers have malpractice, what they think about the severity of the different situations that might be considered misconduct and their conditioning reasons. The results of this study contribute to a better understanding of the extent of research misconduct in 5 Portuguese biomedical institutions, to the perception of which malpractices are considered more serious by researchers, as well as their conditioning reasons. By knowing what actually constitutes research misconduct we will be able to better evaluate the problem, and identifying the reasons for misconduct would make possible to implement measures to prevent them.
Acknowledgments
We thank to the institutions that agreed to participate in the study and researchers who voluntarily completed the questionnaire.
We also acknowledge helpful contribution of Luciano Alves in the development of the electronic version of the questionnaire.
Author Contributions
All authors contributed significantly in the design of the project, development of the questionnaire, analysis of data and manuscript writing.
Competing Interests
The authors disclose any financial and personal relationships with other people or organizations that could inappropriately influence their work.
References
1 Dresser R. Defining research misconduct: will we know it when we see it? The Hastings Center Report. 2001;31(3):31-2. PMID: 11478122.
2 Steneck NH. Assessing the Integrity of Publicly Funded Research. In Investigating Research Integrity. Proceedings of the first ORI Research Conference on Research Integrity, edited by Nicholas H Steneck and Mary D Scheetz, Washington, DC: Office of Research Integrity, 2002, pp 1-16. http://www-personal.umich.edu/ ~nsteneck/publications/Steneck_N_02.pdf (accessed 19 May 2014)
3 Steneck NH. Fostering integrity in research: definitions, current knowledge, and future directions. Science and Engineering Ethics. 2006;12(1):53-74. PMID: 16501647
4 Broome ME, Pryor E, Habermann B, et al. The Scientific Misconduct Questionnaire--Revised (SMQ-R): validation and psychometric testing. Accountability in Research. 2005;12(4):263-80. PMID: 16578917
5 Davis MS, Riske-Morris M, Diaz SR. Causal factors implicated in research misconduct: evidence from ORI case files. Science and Engineering Ethics. 2007;13(4):395-414. PMID: 18038194
6 Smith R. What is research misconduct? In: White C, editor. The COPE Report 2000. London: BMJ Books; 2000. p. 7-11.
7 Wager E, Kleinert S. Cooperation between research institutions and journals on research integrity cases: Guidance from the committee on publication ethics. Saudi Journal of Anaesthesia. 2012;6(2):155-60. PMID: 22754443
8 Lee J. The Past, Present, and Future of Scientific Misconduct Research: What Has Been Done? What Needs to be Done? The Journal of the Professoriate. 2011: 67-83. ISSN 1556-7699. http://jotp.icbche.org/2012/6_1_Lee_67_finalBBJ.pdf (accessed 19 May 2014).
9 Martinson BC, Anderson MS, de Vries R. Scientists behaving badly. Nature. 2005;435(7043):737-8. PMID: 15944677
10 Fanelli D. How many scientists fabricate and falsify research? A systematic review and meta-analysis of survey data. PloS One. 2009;4(5):e5738. PMID: 19478950
11 Steen RG. Retractions in the scientific literature: is the incidence of research fraud increasing? Journal of Medical Ethics. 2011;37(4):249-53. PMID: 21186208
12 International Committee of Medical Journal Editors. Uniform requirements for manuscripts submitted to biomedical journals: Writing and editing for biomedical publication. J Pharmacol Pharmacother. 2010;1(1):42-58. PMID: 21808590
13 Evelyne D, Géraldine S, Laure H. Retractions due to errors and frauds. Presse Med. 2012;41(9 Pt 1):847-52. PMID: 22841376
14 Van Noorden R. Science publishing: The trouble with retractions. Nature. 2011;478(7367):26-8. PMID: 21979026
15 Grieneisen M, Zhang M. A Comprehensive Survey of Retracted Articles from the Scholarly Literature. PloSone. 2012;7(10):e44118. PMID: 23115617
16 Stretton S, Bramich NJ, Keys JR, et al. Publication misconduct and plagiarism retractions: a systematic, retrospective study. Current Medical Research and Opinion. 2012;28(10):1575-83. PMID: 22978774
17 Street JM, Rogers WA, Israel M, et al. Credit where credit is due? Regulation, research integrity and the attribution of authorship in the health sciences. Social Science & Medicine. 2010;70(9):1458-65. PMID: 20172638
18 Budd JM, Sievert M, Schultz TR, et al. Effects of article retraction on citation and practice in medicine. Bulletin of the Medical Library Association. 1999;87(4):437-43. PMID: 10550028
19 Budd JM, Coble ZC, Anderson KM. Retracted Publications in Biomedicine: Cause for Concern. 2011: 390-395. https://surgery.med.uky.edu/sites/default/files/retracted_publicatio.pdf (accessed 19 May 2014)
20 Benos DJ, Fabres J, Farmer J, et al. Ethics and scientific publication. Advances in Physiology Education. 2005;29(2):59-74. PMID: 15905149
2. PROPOSTA DE PROJETO DE INVESTIGAÇÃO
Resumo
Introdução e objetivos: A investigação constitui o principal motor para o progresso
científico. Durante muito tempo assumiu-se que esta atividade era suficientemente auto-regulada e que as más práticas eram muito raras. Porém, com a descoberta de alguns casos de más condutas nos anos 80, este tema foi sendo alvo de crescente preocupação e investigação no seio da comunidade científica internacional. Em Portugal, no entanto, não existem, tanto quanto é do nosso conhecimento, estudos que abordem esta questão numa perspetiva global e abrangente. Neste estudo propomos avaliar a questão da Integridade em Investigação, identificando as más práticas, a gravidade atribuída pelos investigadores às mesmas e os motivos que, na sua opinião, determinam a sua existência. Uma vez identificados os aspetos acima mencionados, é também nosso objetivo propor eventuais medidas para a sua correção.
Material e métodos: Serão dirigidos questionários aos investigadores das instituições
participantes, através do correio electrónico interno das mesmas. Os investigadores terão acesso a um texto introdutório onde serão explicados os principais objetivos do projeto e a um link através do qual poderão proceder ao preenchimento do inquérito. Os dados daqui resultantes serão armazenados numa base de dados (folha Excel) de forma anónima. Para o tratamento estatístico dos mesmos será utilizado o programa SPSS. Será feita uma análise estatística descritiva, de forma a caraterizar a população respondedora do ponto de vista sócio-demográfico (idade, género, experiência prévia no campo da investigação) e calcular, entre outros, as frequências absolutas e relativas das diversas opções de resposta a cada questão, na população como um todo.
Introdução
A investigação constitui o principal motor para o progresso científico, resultando na publicação de inúmeros artigos anualmente. Durante muito tempo foi assumido que esta atividade era suficientemente auto-regulada e que as más práticas eram muito raras. Porém, a partir de 1980, com a descoberta de alguns casos de más condutas, esta crença começou a ser abalada (1, 2). A partir dessa altura, a integridade dos investigadores e as más práticas em investigação têm sido temas com crescente atenção e discussão no seio da comunidade científica (3-5).
Definições
Foram várias as definições de “más práticas em investigação” propostas por diferentes organizações, no entanto nenhuma se mostrou totalmente satisfatória (6, 7). Porém, após um extenso debate, as condutas investigacionais foram divididas em três categorias: más condutas, condutas questionáveis e condutas responsáveis (3).
Uma conduta de investigação responsável é aquela “em que os investigadores cumprem as responsabilidades profissionais definidas pelas organizações a que pertencem, instituições onde trabalham e, quando aplicável, pelo governo e pelo público”(3). Pelo contrário, as más condutas podem ser definidas como “fabricação, falsificação ou plágio na proposta, execução ou revisão de atividades de investigação, ou na publicação dos seus resultados” (8). A fabricação refere-se à “invenção de dados e/ou resultados” (8). A falsificação constitui a “manipulação de materiais, equipamentos ou métodos de investigação e modificação ou omissão de dados ou resultados”(8). Por fim, o plágio consiste na “apropriação indevida de ideias, métodos, resultados ou palavras de outrem”(8). Para além destas, o relatório de 2000 do Committee on Publication Ethics considera também más práticas não obter aprovação da comissão de ética local, não admitir falta de dados, ocultar dados ou resultados discrepantes, não incluir os dados dos efeitos laterais em ensaios clínicos, não obter o consentimento informado em investigação com humanos, elaborar análises
post hoc sem declarar que o são, autorias honorária ou fantasma, publicação redundante, não
declarar conflitos de interesses, não publicar os resultados de um trabalho terminado e não fazer a pesquisa adequada antes de iniciar uma nova investigação (6). Entre estas duas categorias, com uma definição mais difícil de encontrar, situam-se as condutas de investigação questionáveis, as quais “violam os valores tradicionais da investigação e podem ser prejudiciais para o seu processo, mas não deterioram de forma direta a sua integridade” (9). É importante que haja a elaboração e adoção de definições comuns para que seja possível avaliar, responder e mudar os comportamentos em investigação de uma forma uniforme (3).
Prevalência das más práticas em investigação
Os dados sobre as más práticas em investigação podem provir dos casos identificados e investigados, de questionários aplicados aos investigadores do tipo self-report e da análise dos motivos pelos quais os artigos denunciados são retirados das bases de dados. No entanto, nenhum destes métodos é preciso na detecção do seu número exato (10).
Estudos gerais
Os questionários podem ser um bom método para comparar o número de casos que são observados e o número de casos que são efetivamente denunciados, pois o primeiro geralmente é bastante superior ao segundo (3, 10). No entanto, podem sobrestimar os casos de más práticas observados devido ao fato de mais do que um investigador poder estar a indicar o mesmo caso (3).
Já foram realizados vários inquéritos no sentido de identificar a prevalência de más práticas entre os investigadores. Um estudo de Wells et al, em 2008, revelou que apenas 36% das más práticas observadas pelos investigadores principais foram reportadas (11). Um inquérito eletrónico realizado aos autores dos artigos adicionados à Cochrane Database of Systematic Reviews entre 1998 e 2001, permitiu apurar que 5% dos respondentes fabricaram dados ou resultados num estudo no qual participaram nos últimos 10 anos e 17% observaram esta prática no mesmo período (12). Em 2000, outro questionário foi endereçado aos membros do International Society of Clinical
Biostatistics, verificando-se que 51% dos respondentes tinham conhecimento de pelo menos um
caso de más práticas graves no últimos 10 anos e 31% tinham estado num projeto com estas práticas na mesma década (13). Posteriormente, em 2005, Martinson conduziu um inquérito aos investigadores financiados pelo National Institutes of Health, e estes admitiram que já tinham falsificado ou fabricado dados (0.3%), plagiado ideias de outras pessoas (1.4%), omitido dados que contradiziam trabalhos prévios (6%), omitido a utilização de dados insuficientes ou conclusões duvidosas de outros investigadores (12.5%), publicado os mesmos dados/resultados em duas ou mais publicações (4.7%), aceite ser autor inapropriadamente (10%), usado desenhos de investigação inapropriados (15.3%) e feito registos inadequados dos dados/resultados (27.5%) (14).
Finalmente, em 2009, foi realizada a primeira meta-análise de estudos sobre este assunto. Esta permitiu chegar à conclusão que, em média, 2% dos investigadores admitem ter fabricado, falsificado ou modificado dados ou resultados pelo menos uma vez e 33% admitem ter praticado outras práticas de investigação questionáveis. Em relação a comportamentos observados nos colegas, a fabricação, a falsificação e a modificação foram testemunhadas por 14% dos respondentes e outras práticas questionáveis por 72% (15).
Autoria indevida
O número de artigos publicados e o seu impacto na comunidade científica (avaliado pelo número de citações) são usados para avaliar a produtividade e influenciam o prestígio, a progressão na carreira e o financiamento de projetos de investigação. Assim, a autoria inapropriada tanto pode beneficiar como prejudicar indevidamente os investigadores (16-18).
Várias associações internacionais propuseram critérios para a atribuição de autoria de artigos científicos (17). Segundo o International Committee of Medical Journal Editors (ICMJE), os autores devem reunir simultaneamente as seguintes caraterísticas: contribuição substancial na conceção ou desenho do projeto e aquisição de dados ou análise dos mesmos; elaboração do artigo ou revisão crítica do conteúdo intelectual do mesmo e, por fim, aprovação da versão final previamente à publicação (19). Porém, à semelhança do que acontece com as definições apresentadas anteriormente, também aqui não existe uniformidade (20, 21). Em 2007, Wager et al concluíram que 41% das revistas biomédicas não fornecem orientação relativamente aos critérios de autoria, 29% utilizam os critérios do ICMJE e 14% propõem critérios diferentes (20).
Duas das más práticas mais frequentes na atribuição de autoria em artigos científicos são a autoria honorária e a autoria fantasma. Um autor honorário é aquele cuja contribuição para a investigação e escrita do artigo não foi substancial (ex: inclusão de um superior hierárquico como autor apenas para evitar conflitos ou para obtenção das condições logísticas necessárias à realização do estudo), enquanto que um autor fantasma é aquele que contribui substancialmente no desenho do estudo, obtenção e interpretação dos dados e na redacção do artigo, mas não é listado como tal (17).
Muitos têm sido os estudos que refletem preocupações quanto à correta atribuição de autoria (16). Em 1998, Flanagin et al investigaram, através de um questionário aos autores, a questão da autoria em artigos de revisão de 6 revistas médicas. Destes, 26% revelaram ter evidências de autoria honorária e 10% de autoria fantasma (22). O mesmo inquérito foi utilizado por Mowatt et al para estudar a autoria das revisões sistemáticas publicadas na The Cochrane Library. De entre os artigos publicados nos volumes 1 e 2 de 1999, 39% revelaram evidência de autoria honorária e 9% de autoria fantasma (23). Mais recentemente, outro estudo mostrou que a percentagem de presumível autoria honorária nos artigos com 2 ou mais autores publicados em 3 revistas de ciências farmacêuticas foi de 14.3%, e de autoria fantasma foi de 0.9% (24); a proporção de artigos com indícios de autoria honorária foi maior em artigos originais e foi diretamente proporcional ao número de autores. Por fim, Wislar et al apuraram, em 2011, em 6 das revistas biomédicas com maior fator de impacto, uma prevalência de evidência de autoria honorária e/ou fantasma de 21%, ocorrendo estas com frequência maior nos artigos originais comparativamente aos artigos de revisão
(25). Resumindo, e reunindo os resultados de vários estudos, estima-se que a percentagem de autoria honorária varie entre 9 e 60%, e a de autoria fantasma entre 9 e 11% (3).
Outro fenómeno observado nas últimas décadas é a tendência para o aumento no número de autores (18, 26). A pressão exercida sobre os investigadores para a publicação, traduzida pela expressão “publish or perish”, foi um dos motivos apontados quer para o aumento do número de autores por artigo, quer para o aumento de autorias indevidas (17, 26). Segundo Drenth et al, em 20 anos, o principal fator para este aumento no British Medical Journal foi o aumento de autores sénior (27). Porém, este facto também se pode dever a uma possível maior complexidade dos estudos e a um trabalho mais cooperativo (26).
Publicação redundante
A publicação duplicada ou redundante consiste na re-publicação de material previamente publicado, de forma integral ou com pequenas modificações (18, 19). Estas são consideradas más práticas porque infringem as leis dos direitos de autor, levam ao desperdício de tempo dos revisores, aumentam desnecessariamente a quantidade de literatura publicada, enfatizam a importância dos achados e podem influenciar as meta-análises (18). Um exemplo destes efeitos é o estudo de Tramer
et al, de 1997, em que se verificou que 17% dos artigos sobre os efeitos anti-eméticos do fármaco
ondansetron e 28% dos dados dos doentes tratados com este fármaco eram duplicados, o que levou a uma sobrestima de 23% na eficácia do medicamento (28).
Erros de citação e referência
Segundo vários estudos, os erros de precisão das citações ocorrem em 30 e 50% dos casos, enquanto que os erros nas referências a outros artigos são encontrados em 10 a 30% (3). Num estudo de 2002, Rastegar et al analisaram os artigos que referiam o artigo de Kitahata et al de 1996 sobre SIDA, tendo verificado que 11% dos artigos transmitiram as conclusões do estudo de forma errada (29). Em 2004, foi investigada a precisão das citações e referências dos artigos de revistas de fisioterapia de 2000 e 2001, das quais 35.9% continham erros (30).
Extracções de artigos das bases de dados bibliográficas
Após a publicação de um artigo científico, a sua extracção da base de dados justifica-se se houver evidência que está associado a uma má conduta na investigação, erro científico, plágio ou violação das guidelines éticas (31). A percentagem de artigos retirados corresponde a 0.02% de todas as publicações, no entanto, entre 1999 e 2009, foi observado um aumento de 10 vezes na taxa anual de extracção (32-35). As revistas com maior fator de impacto são as que apresentam o maior
número de artigos retirados (31, 33, 36), o que pode estar relacionado com a maior visibilidade e a consequente maior facilidade em serem detetados erros e fraudes (36).
Foram analisadas 42 das maiores bases de dados bibliográficas, incluindo artigos de várias áreas científicas. Dos artigos retirados entre 1928 e 2011 devido a más práticas, 47% deveram-se a más práticas na sua publicação, 20% a más práticas ao longo da investigação e 42% devido a dados ou interpretações dos resultados questionáveis (alguns dos artigos reuniram mais do que um dos critérios anteriores) (34).
Num estudo de 2012, Stretton et al analisaram as publicações retiradas da MEDLINE, entre 1966 e 2008, e verificaram que 52,1% dos artigos tinham sido retirados por falsificação ou fabricação e 41.8% por plágio. Os artigos retirados por plágio tiveram uma relação positiva significativa com autores provenientes de países mais pobres, que não estavam escritos em inglês, que não eram originais e que foram publicados em revistas com baixo índice de impacto (37). No mesmo ano, Fang et al revelaram que 67.4% dos artigos retirados da PubMed deveram-se a más práticas (43.4% por fraude, 14.2% por dupla publicação e 9.8% por plágio), sendo que, segundo o autor, as informações incompletas contidas nos avisos de retirada dos artigos tinham levado a uma subestima da contribuição da fraude em estudos anteriores (38).
As retiradas de artigos escritos em inglês da PubMed, entre 2000 e 2010, foram devidas a erros ou motivos desconhecidos em 73.5% dos casos e a fraude em 26.6%, tendo o número de artigos retirados por fraude aumentado ao longo da década (35). No mesmo período e na mesma base de dados, um outro estudo revelou que os estudos retirados por fraude tinham mais autores do que os artigos retirados por erro, tinham sido retirados de forma mais lenta e, em 53% dos casos, tinham sido escritos por primeiros autores de outros artigos retirados (39).
A identificação dos erros deve ser notada através da publicação de uma correcção ou errata e, no caso de se tratar de um qualquer tipo de fraude, a revista deve publicar uma declaração de denúncia (retraction note) (19). No entanto, segundo o que Steen et al apuraram em 2011, 31.8% das publicações retiradas não se fizeram notar de qualquer forma (35). Para além disso, mesmo que os artigos com erros ou fraudes sejam retirados, eles podem continuar a influenciar a comunidade científica. Em 1999, Budd et al verificaram que os artigos retirados da MEDLINE entre 1966 e 1996 eram citados cerca de 2000 vezes após a sua retirada e que apenas 8% da citações referiam essa extracção (40). O mesmo autor analisou os artigos retirados da mesma base de dados entre 1997 e 2009 e concluiu que o número de citações pós retirada era semelhante e que a referência à extracção só estava presente em 4% dos casos (41).
Motivos condicionadores de más práticas em investigação
Tem havido pouca investigação acerca das causas que motivam as más práticas em investigação (5). São propostos vários motivos condicionadores de más práticas na ciência em geral: desejo de prestígio, desejo de provar as convicções, baixa regulação, pressão para obter financiamentos e publicações, falta de tempo e dificuldade em trabalhar em equipa (10). Para além disso, também se deve ter em conta o desejo de progressão na carreira e os conflitos de interesse de natureza financeira, profissional ou pessoal, entre outros (4). A investigação dos casos de más condutas por estagiários investigados pelo Office of Research Integrity, revelou que 50% destes estavam sob pressão por parte dos tutores, dos quais 75% não acompanharam o processo de obtenção dos dados (42). Em 1992, Kalichman et al inquiriram estudantes da área biomédica e obtiveram 36% de respostas positivas para a observação de algum tipo de más práticas e 15% para a possibilidade dos próprios respondentes as praticarem se isso lhes possibilitasse um financiamento ou a publicação de um artigo (43).
Mais investigação nesta área deve ser realizada para serem apurados de forma mais consistente os motivos condicionadores de más práticas em investigação.
Em Portugal, foi realizado recentemente um estudo (44) sobre a integridade académica dos estudantes universitários portugueses. No entanto, não existem, tanto quanto é do nosso conhecimento, estudos que abordem a questão da integridade em investigação numa perspetiva global e abrangente, desde a conceção e redação das propostas de investigação até à publicação dos resultados, passando pela execução dos projetos e pela análise e interpretação dos dados. São estas as questões que pretendemos abordar neste estudo, que visa identificar as más práticas em investigação, a gravidade atribuída às mesmas e os motivos que determinam a sua existência, assim como propor medidas para a sua correcção.
Problemas
Em Portugal, não existem, tanto quanto é do nosso conhecimento, estudos que abordem a questão da integridade em investigação numa perspetiva global e abrangente. A inexistência de dados sobre as más práticas em investigação é um obstáculo para a implementação de medidas que as tentem minimizar.
Questões
Qual é a prevalência de más práticas em investigação em centros hospitalares e instituições de ensino superior e de investigação do Norte do país (distrito do Porto)?
Quais são os motivos condicionadores de más práticas em investigação?
Objetivos do estudo
Neste estudo propomos avaliar de uma forma abrangente, através de um questionário dirigido aos investigadores de algumas instituições de investigação portuguesas da região Norte, a questão da Integridade em Investigação, identificando as más práticas, a gravidade atribuída pelos investigadores às mesmas e os motivos que, na sua opinião, determinam a sua existência. Uma vez identificados os aspetos acima mencionados, é também nosso objetivo propor eventuais medidas para a sua correção.
Intervenientes
Instituições, Departamentos e Serviços Proponente
Centro Hospitalar do Porto (CHP)
Participantes
Serão convidadas a participar 9 instituições públicas da região Norte do País, distrito do Porto, ligadas à área da Saúde e às ciências médicas e biomédicas: 3 centros hospitalares, 3 instituições de ensino superior e 3 instituições de investigação:
Instituições hospitalares
o Centro Hospitalar do Porto (CHP) o Centro Hospitalar de São João (CHSJ)
o Centro Hospitalar de Vila Nova de Gaia / Espinho (CHVNG/E) Instituições de ensino superior
o Instituto de Ciências Biomédicas Abel Salazar da Universidade do Porto (ICBAS/UP)
o Faculdade de Farmácia da Universidade do Porto (FF/UP) o Faculdade de Medicina da Universidade do Porto (FM/UP) Instituições de investigação
o Instituto de Biologia Molecular e Celular da Universidade do Porto (IBMC/UP) o Instituto de Engenharia Biomédica da Universidade do Porto (INEB/UP)
o Instituto de Patologia e Imunologia Molecular da Universidade do Porto (IPATIMUP)
Equipa de Investigação Constituição
Aluno
Cláudia Silva: aluna da Disciplina de Iniciação à Investigação Clínica (DIIC) do Curso de Mestrado Integrado em Medicina (MIM) do ICBAS/UP
Orientadores do projeto
Margarida Lima: médica, assistente hospitalar graduada de Imunohemoterapia do Serviço de Hematologia Clínica do CHP; doutorada em Ciências Médicas pelo ICBAS/UP; professora auxiliar convidada do ICBAS/UP; diretora do Departamento de Ensino, Formação e Investigação (DEFI) do CHP; regente da DIIC.
Supervisor da DIIC
Margarida Lima
Outros investigadores
Isabel Fonseca: nutricionista, técnica superior de saúde, do Serviço de Nutrição e Alimentação do CHP; mestre em Saúde Pública – especialidade de Bioestatística; aluna
